Multi-bit, monotonic quantizer and linearized delta-sigma modulator based analog-to-digital converter (ADC) modulators and digital-to-analog converter (DAC) modulators are disclosed. A folded monotonic quantizer provides tri-state to bi-state mapping such that three state bits may be utilized but are not required to be transported and stored. The multi-bit ADC and DAC modulators provide linearized, spurious free outputs that require less power and utilize lower oversampling ratios to provide wider bandwidths at high-performance, spurious-free dynamic ranges, making them ideally suitable for use in direct sampling receivers and direct digital transmitters.
|
15. An analog-to-digital converter, comprising:
means for receiving an analog input signal; means for quantizing the analog input signal into a tri-state signal at a predetermined sample rate; and means for mapping the tri-state signal into a bi-state signal whereby an output signal is provided, the output signal being represented by a predetermined number of bits.
10. A digital-to-analog converter modulator, comprising:
means for modulating a digital input signal being represented by a predetermined number of bits; means for mapping an output of said modulating means from a two state signal to a three state signal; and at least one or more means for converting a three state signal output from said mapping means from a digital signal to an analog signal wherein one converting means is provided per each of the predetermined number of three state bits.
1. An analog-to-digital converter modulator, comprising:
means for modulating an analog input signal; means for quantizing an output received from said modulating means; means, coupled to said quantizing means, for mapping an output of said quantizing means from a three state code to a two state code; means, coupled to said mapping means, for delaying an output of said quantizing means; and means for providing a linearized feedback signal from said delaying means to said modulating means wherein an analog signal is converted to a digital signal having enhanced linearity and being represented by a predetermined number of bits.
2. An analog-to-digital converter modulator as claimed in
3. An analog-to-digital converter modulator as claimed in
4. An analog-to-digital converter modulator as claimed in
5. An analog-to-digital converter modulator as claimed in
6. An analog-to-digital converter modulator as claimed in
7. An analog-to-digital converter modulator as claimed in
8. An analog-to-digital converter modulator as claimed in
9. An analog-to-digital converter modulator as claimed in
11. A digital-to-analog converter modulator as claimed in
12. A digital-to-analog converter modulator as claimed in
13. A digital-to-analog converter modulator as claimed in
14. A digital-to-analog converter modulator as claimed in
16. An analog-to-digital converter as claimed in
18. An analog-to-digital converter as claimed in
19. An analog-to-digital converter as claimed in
20. An analog-to-digital converter as claimed in
|
The present invention generally relates to the field of data converters, and particularly to multi-bit, high performance analog-to-digital and digital-to-analog conversion.
Delta-sigma modular architectures for analog-to-digital converters (ADCs) and digital-to-analog converters (DACs) based on continuous-time and switched capacitor integration are well known in the art. These architectures typically are for one bit or more modulators, one bit being the most common since the quantizers and DACs in the modulators are inherently more linear for one bit. The one bit minimizes the major sources of distortion generation in the modulator and therefore allows for high levels of spurious free dynamic range. However, a high-performance modulator of order two or more integrator stages is only conditionally stable and must incorporate substantial performance loss to maintain stability. This is especially true for the one bit modulator, since the two state output cannot instantaneously track the continuous, band limited (filtered) input signal. It is only over the band limited longer term that the output tracks the input precisely. If, on the other hand, multi-bit modulators are used, the output quantization noise level is lower to begin with, and the instantaneous output is able to more closely track with the input. This closer tracking equates to less performance loss that must be incorporated into multi-stage modulators to maintain stability. In fact, most two stage multi-bit modulators do not have to purposely add loss in order to achieve stability since sufficient loss in the actual circuit is already present. Therefore, multi-bit modulators would be the best approach for high-performance ADCs and DACs were it not for the difficulty in maintaining linearity so as to maintain the desired high level of spurious free dynamic range. Since multi-bit modulators are much more efficient in suppressing in-band noise, the oversampling ratio (sampling rate divided by twice the resolution bandwidth) can be lower, which generally equates to lower power required and/or less costly parts. It would therefore be highly desirable to provide a high-performance, multi-bit ADC or DAC modulator that provides good linearity and high spurious free dynamic range.
The present invention provides a means for acquiring and maintaining linearity in high-performance ADCs and DACs. Thus, the present invention is directed to an analog-to-digital converter. In one embodiment, the delta-sigma analog-to-digital converter modulator includes a means for modulating an analog input signal, a means for quantizing an output received from the modulating means to a three state digital code, a means for mapping an output of the quantizing means from a three state digital code to a two state digital code, a means for delaying and reordering the two state digital code output by a predetermined or random means, a means for inverse mapping the reordered two state digital code to a three state digital code, and a means for converting the resultant three state digital code to an analog signal which is connected back to the modulating means. The analog-to-digital converter thus formed is capable of converting an analog input signal to a two state digital output code with a predetermined number of bits of resolution.
The present invention is further directed to a digital-to-analog converter. In one embodiment, the delta-sigma digital-to-analog converter modulator includes a means for modulating a digital input signal being represented by a predetermined number of bits, a means for mapping an output of the modulating means from a two state signal to a three state digital code, one means per bit of the three state digital code output for converting each bit to an analog signal, and a combiner and reconstruction filter means for combining the analog signals. The digital-to-analog converter thus formed is capable of converting a digital input signal to an analog output signal at the output of the reconstruction filter means.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention as claimed.
The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention and together with the general description, serve to explain the principles of the invention.
The numerous advantages of the present invention may be better understood by those skilled in the art by reference to the accompanying figures in which:
Reference will now be made in detail to the presently preferred embodiment of the invention, an example of which is illustrated in the accompanying drawings.
Referring now to
ADC modulator 100 consists of an N-stage delta-sigma (low pass or bandpass) modulator filter 112 driving a folded M-bit quantizer 114, the output of which has three states per bit and is mapped to M two state bits with 3 state to 2 state map 116. After delaying the bit set by one clock delay by register delay set 118, the resampled bits are ready to be sent to a suitable digital signal decimation circuit to recover the desired binary bits of resolution and reduce the output sample rate. Sample rate reduction demultiplexer circuit 120 may be optionally utilized prior to decimation filtering to reduce the data output bus rate at modulator output 122 (serial-to-parallel bit set regrouper). This function would accommodate slower, parallel processed decimation filters that are not capable of processing the output data directly at the serial sample output rate.
A feedback signal is formed by mapping the two state bit output data from register delay set back to 3 state bit groups with 2 state to 3 state map 132 and applying a three state DAC 126, 128 and 130 to each bit (one DAC per bit). The analog output of each of DACs 126-130 is either +V, -V, or off (ground), depending upon the corresponding input data states. All bits are of equal weighting and can be combined by simply summing the outputs of DACs 126-130 with summer 124. In order for the noise and distortion in the data output of ADC modulator 100 to be minimized, the outputs of DACs 126-130 are preferably identical in span, and summer 124 preferably provides equal unity weight for each bit. Thus, summer 124 is preferably a unity weight summer. To compensate for mismatches, a calibrate on demand mode may be provided whereby the data into DACs 126-130 can be set to known patterns that result in null conditions (ground) in the summed output provided by summer 124. The individual values of +V and -V output levels of each of DACs 126-130 may be adjusted with DAC output gain calibration controller 134 to establish the desired balance between the bits. The DAC output gain calibration controller 134 determines desired DAC outputs by observation of null conditions at the output of unity weight summer 124. The calibration process balances out both DAC and summing input imbalances. Calibration mode may be controlled with a calibration mode input 136. If there is still undesired imbalance between the bits, precluding meeting the desired level of linearity, this small remaining imbalance can be randomized and distributed equally over all the bits. This is done with an optional cross distribution switch set 138 shown, which randomly scrambles the bits in the feedback with random selection controller 140 and converts non-linearity residue to a slight increase in the output noise floor of ADC modulator 100.
Referring now to
One common way of implementing a two's complement quantizer is by using a "thermometer code" output stack of comparators and a thermometer code to two's complement code converter. This ADC approach is commonly called a "flash ADC converter". Because the output is a binary code, the relationship between the code states and the voltage thresholds that produced the thermometer code must be precise and uniform or non-linearity produced distortion in the two's complement output will result. This means that the thresholds used in the thermometer code must be the same incremental voltage apart. If the code converter were left off and just the thermometer code output were used, the code can be considered as a single sided monotonic code formed from two state bits (e.g., ON state and OFF state).
The thresholds would not have to be precisely spaced as long as quantizer 114 is used with delta-sigma modulator 112 which can suppress the distortion that would normally result from any non-uniformity. It is quite difficult to establish and maintain a precise, uniformly space comparator set, so some means of suppressing the resulting distortion (such as a delta-sigma modulator) is preferred for high spurious free dynamic range. Further, a two sided monotonic code formed from three state bits is desired for utilization in a delta-sigma modulator in the event that plus (+) and minus (-) voltage signals are desired.
A diagram of a 4-bit monotonic folded quantizer 218 provided in the invention which uses four bits to quantize eight states is shown in
By using amplitude folding, the same threshold pair values can be used for every bit as shown in quantizer signal diagram 228. The folding is done simply by providing different gain adjustments (G1, G2, G3, etc.) 240, 242 and 244, respectively, in the signal path 230 against a common set of +/-thresholds, to form uniformly spaced, effective threshold points over the signal as shown at 210 and 218. Although the gain adjustments need not be precise when quantizer 114 is part of a delta-sigma modulator, an automatic gain control scheme could be considered to maintain uniformity if quantizer 114 were to be used without a delta-sigma modulator 112. Since the thresholds are all the same, they can be formed by whatever means is the most accurate. For example, a full-wave rectification scheme could be used with a single threshold rather than a threshold pair to determine when the bit is to change, and then a +V or -V selection based on the polarity of the original signal can be used to select the output state. In this example, the positive and negative thresholds are matched because they are implemented as the same threshold (assuming that the full-wave ratification is near ideal and does not add distortion to the signal).
The monotonic bits formed are all equal weight and the coded value is formed by simply adding the bits (i.e., as shown by folded quantizer 218 and bit output signals 220-226, adding the bits as the waveforms are traversed from left to right (+1, +2, +3, +4, -4, -3, -2 and -1). However, it is advantageous to not have to transport and store three state bits. Fortunately, the monotonic bit structure used can be mapped to two state bits for storage and transport and then mapped back to be processed further. By observing that LSB 220 always has two states, and that the other bits only have two states relative to LSB 220, the bits can be mapped as shown by bit waveforms 254, 256, 258 and 260. That is, if LSB 220 is positive, the other bits are either OFF or positive. Likewise, if LSB 220 is negative, the other bits are either OFF or negative. OFF state bits can therefore be mapped to the opposite state from LSB 220, and recognized as such when mapping back.
The number of monotonic quantizer bits required can be generalized as ½ the number of quantizer code states. Since the binary code (two's compliment code) bits required is log2(g), where g is number of quantizer code states, the monotonic code can provide finer resolution per bit but at the expense more bits per quantizer state set. The following table shows the comparison:
TABLE 1 | ||
Comparision between code states and two's complement and | ||
montonic bits | ||
States | 2s Complement Bits | Monotonic Bits |
2 | 1 | 1 |
4 | 2 | 3 |
6 | -- | 3 |
8 | 3 | 4 |
10 | -- | 5 |
12 | -- | 6 |
14 | -- | 7 |
16 | 4 | 8 |
18 | -- | 9 |
20 | -- | 10 |
22 | -- | 11 |
24 | -- | 12 |
26 | -- | 13 |
28 | -- | 14 |
30 | -- | 15 |
32 | 5 | 16 |
Referring now to
A 3-state signal is converted into a 2-state signal according to the following rules. If the LSB is equal to -V, then the LSB is set to -V, otherwise the LSB is set to +V. If the MSB is equal to -V, or the LSB =+V, then the MSB is set to -V, otherwise the MSB is set to +V. If the middle bit is equal to the LSB which is equal to -V, or the LSB is equal to 0, then the middle bit is set to -V, otherwise, if the middle bit is equal to the LSB which is equal to +V, then the middle bit is set to +V, otherwise if the LSB is equal to -V, then the middle bit is set to +V, otherwise, the middle bit is set to -V.
A 2-state signal is converted to a 3-state signal according to the following rules. If the LSB is equal to the MSB which is equal to +V, then the LSB is set to 0, otherwise the LSB is set to the present value of the LSB. If the LSB or the MSB is equal to +V, then the MSB is set to 0, otherwise, the MSB is set to the present value of the MSB. If the middle bit is not equal to the LSB, then the middle bit is set to 0, otherwise, the middle bit is set to the present value of the middle bit.
Referring now to
The multi-bit DAC 400 shown in
Assuming that the DAC output 434 is worst case peak to peak symmetrical (i.e. two's complement -1 is not used), Table 2 shows example two's complement to tri-state bit mappings. The mappings can be determined for higher bit resolutions and for full (non-symmetrical) two's complement notation by using the equations shown below.
TABLE 2 | |||||
Two's complement to tri-state bit mapping. | |||||
2-Bit-2's | |||||
Complement to 1 | 3 Bit-2's Complement to | 4 Bit-Two's Complement to | |||
Bit Tri-State Map | 3 Bit Tri-State Map | 7 Bit Tri-State Map | |||
01 | 1 | 011 | 111 | 0111 | 1111111 |
00 | 0 | 010 | 011 | 0110 | 0111111 |
11 | -1 | 001 | 001 | 0101 | 0011111 |
000 | 000 | 0100 | 0001111 | ||
111 | 00-1 | 0011 | 0000111 | ||
110 | 0-1-1 | 0010 | 0000011 | ||
101 | -1-1-1 | 0001 | 0000001 | ||
0000 | 0000000 | ||||
1111 | 000000-1 | ||||
1110 | 00000-1-1 | ||||
1101 | 0000-1-1-1 | ||||
1100 | 000-1-1-1-1 | ||||
1011 | 00-1-1-1-1-1 | ||||
1010 | 0-1-1-1-1-1-1 | ||||
1001 | -1-1-1-1-1-1-1 | ||||
For the data shown in Table 2, the data is symmetric about zero (neglecting -1of two's complement representation). With M two's complement bits in, there are P tri-state bits out where P=(2(M-1)-1. If the full two's complement number space is required (i.e. includes -1), then P=(2(M-1)). Each P tri-state bit utilizes a separate 3 state DAC. As is typical with a DAC, the anolog output is filtered with a reconstruction filter to remove the sampled spectrum artifacts above the desired bandwidth. Thus, DAC 400 preferably includes a unity weight summer 430 to combine the individual component (bit) DACs and a reconstruction filter 432, and is also utilized to suppress out-of-band noise. In the event the delta-sigma modulator is a bandpass form, the signal bandwidth has shaped noise at higher and lower out-of-band frequencies. As such, reconstruction filter 432 would be a bandpass filter in the event DAC 400 is a a bandpass DAC.
It is believed that the analog-to-digital and digital-to-analog converters of the present invention and many of their attendant advantages will be understood by the foregoing description, and it will be apparent that various changes may be made in the form, construction and arrangement of the components thereof without departing from the scope and spirit of the invention or without sacrificing all of its material advantages. The forms herein before described are merely explanatory embodiments thereof. It is the intention of the following claims to encompass and include such changes.
Patent | Priority | Assignee | Title |
10411726, | Dec 14 2017 | Samsung Electronics Co., Ltd. | Quantizer including capacitors and operating method of quantizer |
6680680, | Aug 16 1999 | STMICROELECTRONICS S A | Process and device for converting an analog signal into a digital signal with automatic gain control |
6765518, | Jul 01 1999 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for efficient mixed signal processing in a digital amplifier |
6791404, | Jul 01 1999 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for efficient mixed signal processing in a digital amplifier |
6864815, | Jul 01 1999 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for efficient mixed signal processing in a digital amplifier |
6894632, | Jan 14 2004 | Northrop Grumman Systems Corporation | Programmable analog-to-digital converter |
6933778, | Jul 01 1999 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for efficient mixed signal processing in a digital amplifier |
6933870, | Jul 01 1999 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for efficient mixed signal processing in a digital amplifier |
6963297, | Nov 12 2003 | Northrop Grumman Systems Corporation | Delta-sigma digital-to-analog converter assembly |
7009543, | Jan 16 2004 | Cirrus Logic, Inc.; CIRRUS LOGIC INC | Multiple non-monotonic quantizer regions for noise shaping |
7084797, | Oct 28 2003 | PANASONIC SEMICONDUCTOR SOLUTIONS CO , LTD | Delta sigma modulating apparatus |
7084800, | Aug 19 2003 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | System and method for shuffling mapping sequences |
7098833, | Jun 04 2004 | Texas Instruments Incorporated | Tri-value decoder circuit and method |
7525455, | Mar 04 2002 | STMICROELECTRONICS N V | Coder apparatus for resonant power conversion and method |
7593483, | May 07 2004 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Nonlinear mapping in digital-to-analog and analog-to-digital converters |
7924937, | Mar 04 2002 | STMICROELECTRONICS N V | Resonant power converter for radio frequency transmission and method |
8018363, | May 07 2004 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Nonlinear mapping in digital-to-analog and analog-to-digital converters |
8233412, | Jun 30 2010 | Rockwell Collins, Inc. | Multi-signal, software-defined and staring cognitive communications system |
8269661, | Oct 14 2010 | Texas Instruments Incorporated | Pipelined ADC having a three-level DAC elements |
8340224, | Mar 04 2002 | STMicroelectronics N.V. | Resonant power converter for radio frequency transmission and method |
Patent | Priority | Assignee | Title |
5245343, | Aug 03 1990 | HONEYWELL INC , A DELAWARE CORP | Enhanced accuracy delta-sigma A/D converter |
6061008, | Dec 19 1997 | HANGER SOLUTIONS, LLC | Sigma-delta-sigma modulator for high performance analog-to-digital and digital-to-analog conversion |
6124813, | Jun 06 1997 | Analog Devices, Inc | Self-linearizing multi-bit DACs |
6188345, | Mar 30 1998 | Texas Instruments Incorporated | High speed delta-sigma A/D converter using a quantized USM negative feedback loop |
6229466, | Aug 23 1999 | LEVEL ONE COMMUNICATIONS, INC | Digital calibration method and apparatus for multi-bit delta-sigma D/A converter |
6326912, | Sep 24 1999 | AKM SEMICONDUCTOR, INC | Analog-to-digital conversion using a multi-bit analog delta-sigma modulator combined with a one-bit digital delta-sigma modulator |
6373418, | May 25 2000 | Rockwell Collins, Inc.; Rockwell Collins, Inc | Nyquist response restoring delta-sigma modulator based analog to digital and digital to analog conversion |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Aug 26 1999 | ABBEY, DUANE L | Rockwell Collins, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 010211 | /0551 | |
May 02 2001 | TUSAN, ROBERT D | ADC DSL SYSTEMS INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012586 | /0552 | |
May 02 2001 | HERRINGTON, THOMAS A | ADC DSL SYSTEMS INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012586 | /0552 | |
May 02 2001 | ARREOLA, JERRY | ADC DSL SYSTEMS INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012586 | /0552 |
Date | Maintenance Fee Events |
Jan 17 2007 | REM: Maintenance Fee Reminder Mailed. |
Feb 22 2007 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
Feb 22 2007 | M1554: Surcharge for Late Payment, Large Entity. |
Feb 07 2011 | REM: Maintenance Fee Reminder Mailed. |
Feb 09 2011 | M1552: Payment of Maintenance Fee, 8th Year, Large Entity. |
Feb 09 2011 | M1555: 7.5 yr surcharge - late pmt w/in 6 mo, Large Entity. |
Jan 01 2015 | M1553: Payment of Maintenance Fee, 12th Year, Large Entity. |
Date | Maintenance Schedule |
Jul 01 2006 | 4 years fee payment window open |
Jan 01 2007 | 6 months grace period start (w surcharge) |
Jul 01 2007 | patent expiry (for year 4) |
Jul 01 2009 | 2 years to revive unintentionally abandoned end. (for year 4) |
Jul 01 2010 | 8 years fee payment window open |
Jan 01 2011 | 6 months grace period start (w surcharge) |
Jul 01 2011 | patent expiry (for year 8) |
Jul 01 2013 | 2 years to revive unintentionally abandoned end. (for year 8) |
Jul 01 2014 | 12 years fee payment window open |
Jan 01 2015 | 6 months grace period start (w surcharge) |
Jul 01 2015 | patent expiry (for year 12) |
Jul 01 2017 | 2 years to revive unintentionally abandoned end. (for year 12) |